F. Bouteldjaoui

You should apply various interpolation techniques and compare the results.

Hello sir,

I have used ArcGIS 10.5 for spatial variation of Drinking Quality Index (DWQI) considering groundwater. And the graphical outcome result was satisfactory. You can also check the following link:

Article Application of GIS in ground water quality mapping for Varanasi city

Article Comparison of interpolation methods for the estimation of gr...

Local polynomial method and Inverse Distance Weighting (IDW) method

Kriging is one of the best methods for this kind of data interpolation.

though the nature of the data used is of great importance for geostatistical analysis, there is no rule of thump to say which model is the best. I suggest you to apply different methods to infer the most suitable one according to the accuracy parameters offered . meanwhile you can also cross check the precision level statistically.

regards

Inverse Distance Weighted Interpolation

depends on the accuracy of your data.

Inverse Distance Weighted (IDW) Interpolation is good for you.

F. Bouteldjaoui

You should apply various interpolation techniques and compare the results.

Nice question

Various method can be used in mapping the spatial distribution of water quality index such as krigging method, inverse distance weighted, local polynomial method etc

As i said in a last post ( "Which are the best interpolation techniques for interpolation of precipitation in Qgis and ArcGIS?") , you can have a quick look to these links :

http://www.gisresources.com/types-interpolation-methods_3/

You can have a quick look to this articles :

http://www.pressesagro.be/base/text/v17n2/392.pdf

Best regards,

HP

I will comment on several interpolation analyzes that are carried out in Argis 10.5 so that the analysis is closest to the physical-geographic characteristics where you obtained the water samples and to the physical, chemical parameters measured for the evaluation of water quality.

Interpolation predicts values for cells in a raster from a limited number of sample data points. It can be used to predict unknown values of any data from a geographic point, such as: elevation, rainfall, chemical concentrations, levels of contamination, etc.

The location of the place where the data is obtained can be a determining factor in the analysis of the data. Ideally, a pattern should be made that offers complete and uniform coverage of the region. However, it should be mentioned that uniform patterns can induce false results if they coincide in their period with some regular phenomenon present in the field. For this reason, data acquisition patterns that offer almost no degree of uniformity are commonly used.

Kriging: The Kriging interpolator uses the characteristics of variability and spatial correlation of the studied phenomenon in the estimation, so its use implies a prior analysis of the information, in order to define or extract from this initial information a model that represents its spatial continuity. This interpolation method assumes that the distance or direction between sample points reflects a spatial correlation that can be used to explain the variation in the surface.

IDW: This interpolation method assumes that each point has a local influence that decreases with distance. In this way, the method weighs the points closest to the process cell with greater force and with less intensity over those located at a greater distance. To define the value of each cell, the program can perform a search based on a defined series of points or it can search for all those points that are within a certain search radius. This method is appropriate when the analyzed variable decreases as the distance increases from each analyzed point. For example, when trying to interpolate a surface that represents the contamination of an industry on the surrounding population. The further away people are from the industry, the less exposure to pollution from the industry will be.

Spline: This generic use interpolation method adjusts a minimum curvature surface through the entered points. It is suitable to show gradual variations of surfaces such as elevations or concentrations of pollution. However, it may not be useful when there are a large number of variations over a relatively small horizontal distance. Spline estimates values using a mathematical function that minimizes the curvature of the total surface, resulting in a smooth surface that passes exactly through the sampled points. This method is best for carefully varying surfaces such as elevation, groundwater surface heights, or concentrations of contamination.

Spline with Barrier: This method is similar to the previous one, but this time its parameters cannot be modified and it also allows the incorporation of barriers (such as cliffs, cavities, or other objects that interrupt the continuity of the analyzed values) which they can be lines or polygons. Spline with Barrier is an interpolation method that uses a method similar to the technique used in the Spline method, but the main difference is that this tool distinguishes encoded discontinuities in both barriers of entry and data from the point of entry. This time its parameters cannot be modified and it also allows the incorporation of barriers (such as cliffs, cavities, or other objects that interrupt the continuity of the analyzed values) which can be lines or polygons.

Natural Neighbors: This generic use interpolation method adjusts a minimum curvature surface through the entered points. It is suitable to show gradual variations of surfaces such as elevations or concentrations of pollution. However, it may not be useful when there are a large number of variations over a relatively small horizontal distance. This method is best suited when values are unevenly distributed. It is an interpolation technique very suitable for generic purposes and has the advantage of not having specific and complex parameters such as radius, proximity, lag, saddle or others.

Trend: This interpolation method fits a mathematical function (polynomial of a user-specified order) to all entered points. Trend uses a least square regression fit that results in a Raster that minimizes the variance of the surface relative to the entered points. This surface is constructed in such a way that, for each interpolation location, the total of the differences between the current values and the estimated values (the variance) will always be as small as possible. In this way, the final result will be a Raster that contains the generalized trend of the values contained in the entered points.

Good luck in your work

I have used IDW and I think it is perfect

There is no best method, you should try defrent methodes and compare their performance (RMSE , MAE, MSE , R^2)

Hello

I believe that we must compare several methods and then see the best, we must also see the variance maps to be able to decide. for my case I compared the IDW with 4 different weights, the Bayesian Kriging, the CoKriging and the Ordinary Kriging and it is the coK then the BK which gave the most favorable results